Large-eddy simulation of a gas turbine combustor with a subgrid model based on a parameterization of the linear-eddy model (LEM) is presented. The combustor chosen for this study involved the General Electric LM6000 lean pre-mixed, dry, low-NOx combustor. The subgrid model was formulated by populating a database of turbulent flame speed statistics from a LEM–counter flow (CF) solver (Calhoon [1]) that was parameterized in terms of subgrid Reynolds and Karlovitz numbers. This combustor was also modeled using the thin-flame formulation of Pocheau [2], which was correlated to match experimental flame speed data for methane–air and ethylene–air premixed flames [3].
The LEM-CF subgrid model formulation shows overall good agreement with the experimental data, and a substantial improvement compared to the correlated Pocheau model. The Pocheau formulation, though tuned to match experimental data at different conditions, predicts a much larger turbulent flame speed than the LEM-CF formulation. The basic LEM was calibrated by Smith and Menon [4] using experimental data for premixed flame speeds at different conditions than for the LM6000 case. However, using these same coefficients, the LEM-CF formulation produced good agreement for the LM6000 case. This demonstrates the superiority of the LEM-CF’s first principles, physics based prediction of the flame properties, and how its formulation may be applied with a much greater degree of confidence, to conditions different from what it has originally been developed for and tested with.
